Piper (Local)

The PiperTTSService provides integration with Piper, a fast, local neural text-to-speech engine. It is designed to run on low-power devices while maintaining high-quality output without any cloud costs.

Installation

1. Install and run a Piper HTTP server (e.g., piper-http).
2. Install the SDK:

   pip install piopiy-ai

Configuration

PiperTTSService Parameters

Parameter	Type	Default	Description
base_url	str	Required	URL of the Piper HTTP server.
aiohttp_session	ClientSession	Required	Shared aiohttp session.
sample_rate	int	None	Output sample rate (defaults to model's native rate).

Usage

Basic Setup

import aiohttp
from piopiy.services.piper.tts import PiperTTSService

async with aiohttp.ClientSession() as session:
    tts = PiperTTSService(
        base_url="http://localhost:5000",
        aiohttp_session=session
    )

Notes

• Model Dependency: The sample rate and voice quality depend entirely on the model file loaded by the Piper server.
• WAV Handling: The service automatically strips WAV headers returned by the Piper server to provide raw PCM data.
• Latency: Being local, Piper offers extremely low latency, often faster than real-time even on basic hardware.